It indeed seems much more logical to associate the Audio_span_dynamic of each note in Dynamic_performer rather than Staff_performer, and avoid the side effects caused by identifying Voices by their name in Staff_performer (in the corner case of identically named Voices).
(I don't really have a strong opinion about whether these side effects could have originally been intentional or not [as sharing dynamics between Voices could even have its uses] since, taken in isolation, the previous Staff_performer code doesn't give any obvious clues about this either way. As there are other developers who think the previous behavior is a bug, however, I have nothing against fixing the behavior like this, and will simplify the documentation patch issue 4920 accordingly.) Having regression tests for this functionality could be useful to document the intended behavior after this change (and might be helpful to anyone trying to understand the design behind the implementation later). https://codereview.appspot.com/304260043/ _______________________________________________ lilypond-devel mailing list lilypond-devel@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-devel